Computers can be used to identify matches between data sets. Current approaches are suboptimal.